Well thank you very much for these explanations. Unfortunately, I must admit
the
book I have for survival analysis seems less precise as to which test to use
and
why.
Still, in coxph (survival), if I have multiple variables in a model, say X_1,
X_2, and X_3, how do I test their respective coefficients \beta_1, \beta_2, and
\beta_3 with the LR, score and Wald? I guess i can do it by comparing the model
with all three variables to those without each of the variables, but is there
not a more straightforward manner?
David Biau.
________________________________
De : "Therneau, Terry M., Ph.D." <thern...@mayo.edu>
À : David Winsemius <dwinsem...@comcast.net>; Biau David <djmb...@yahoo.fr>
Cc : r help list <r-help@r-project.org>
Envoyé le : Ven 30 juillet 2010, 19h 07min 15s
Objet : RE: Re : [R] COXPH: how to get the score test and likelihood ratio test
for a specific variable in a multivariate Coxph ?
The Wald, score, and LR tests are discussed in full in my book. They
are not the same.
The LR test is the difference between LR(beta=0) and LR(beta=final). The
score test is a Taylor series approximation to this using an expansion
around beta=0. The Wald test is a similar Taylor series approximation,
but around beta=final.
If there are no tied times the score test = Log-rank test. If there
are ties, then they are just a tiny bit different: the paper using the
log-rank has an n-1 in his variance term and the Cox model has an n.
Neither is right or wrong, just a different choice.
Terry Therneau
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
